image/svg+xml Isobel Bicket CAS741Prof: Dr S. SmithSeptember 26, 2017 x y E Sample Spectrum What is... A 3D dataset with:2 spatial dimensions1 spectral dimension Originates from a technique such as electron energy loss spectroscopy or cathodoluminescence ...which are what? Electron energy loss spectroscopyis done in a transmission electron microscope Sample Electron beam An beam of electrons isaccelerated towards asample The electrons in the beaminteract with the sample:some may lose energy tothe sample. The transmitted electrons arecollected in a spectrometerand dispersed accordingto energy: an electron energy loss spectrum [Graphic courtesy of Dr T. Coenen] Cathodoluminescencecan be done in a transmission electron microscopeor a scanning electron microscope A beam of electrons interacts with the sample;the sample may emit light A mirror collects emitted lightand sends it to a spectrometer In both cases the beam can be rastered across the sample, and a spectrum is acquired at each pixel. The spectrometer software outputs a *.dm3 (EELS) or *.h5 (CL) file. Goal 1: Read the 3D dataset We consider the 3D dataset as a 'model' of the samplecombined with the instrument response- The instrumental response can broaden peaks (reduce resolution) or add background signal Physical system 1 Physical system 3 Physical system 2 Goal 2: Provide the usera way to navigate the data Goal 3: Remove the instrumentcontribution via normalizationand deconvolution (EELS) or background subtraction (CL) Normalization Normalize intensity (I) of each (x,y) pixel to the integral of the spectrum (E) Normalize intensity (I) of each (x,y) pixel to the integral of a chosen range in the spectrum (E) Normalize intensity (I) of each (x,y) pixel to the maximum value of intensity along the E axis Deconvolution Convolve point spread function (PSF) (P(a/b)) with (measured spectrum D(a)divided by convolution of PSF (P(a/b)) with 'real spectrum' (Oc(d))) and multiply by 'real spectrum' (Oc(b)) to find the next estimation of the 'real spectrum' (Oc+1(b)).'Real spectrum' (Oc) is the best estimation of the true spectrum at iteration c Background subtraction Remove dark noise Remove substrate signal Correct for sensitivity (γspectrometer) of spectrometer ...data definitions?...instance models? ...theoretical models? Energy conservation Interaction probability equations- Energy loss probability- Emission probabilityPoissonian detector noise Goal 4: Export images or spectra as desired by user Assumption: EELS data can be described accurately as convolution of sample response with the system response Assumption: Fluctuations in signal are artifacts due to changes in collected beam current Assumption: Dark/substrate signal can be accurately removed with simple subtractionAssumption: Spectrometer sensitivity is accurately modelled with experimental reference data x y E Sample Spectrum (x,y): imageE: Spectrum Data Constraints Physical:-Maximum intensity depends on spectrometer saturation value (depends on acquisition hardware and software)- Intensity > 0 (ideal, but not always true)- Max(Energy loss) < microscope beam energy (EELS) Software:...? Functional and Non-functional Requirements Accept 3D datasets as inputRead *.dm3, *.h5 filesCorrectly apply artifact correction routines (ie, normalization, deconvolution, background subtraction, sensitivity correction)Allow user to select arbitrary slices along the spectral axis and display (and export) an averaged image (x,y axes)Allow user to select arbitrary areas of an image (x,y axes) and display (and export) a spectrumUseable by a microscopist unskilled in programmingEasily set-up and run on Linux, Windows, MacOSReturn processing results quickly (quantify/qualify?) ?? A software for the data processing of spectrum imagedata resulting from electron energy loss spectroscopyor cathodoluminescence spectroscopy Software Requirements Specification SpectrumImageAnalysisPy What is a... [Titan-LB, CCEM, ABB-B161]
1
  1. Title
  2. What is SIAnPy?
  3. SIAnPy
  4. What is a SI?
  5. A SI
  6. What are EELS and CL?
  7. EELS in a TEM
  8. EELS Physical system 1,2
  9. EELS Theoretical models
  10. CL
  11. Goals 1,2
  12. Goal 3
  13. Normalization
  14. Deconvolution
  15. Background subtraction
  16. DD, IM?
  17. Goal 4
  18. Data Constraints
  19. Data Constraints